The work aims to develop an integrated control mode of cigarette manufacturing quality to solve the low control accuracy, large quality variation and other problems in cigarette manufacturing. The cumulative sum method was used to predict the deviation of weight and circumference quality characteristics and adjust online. At the same time, with the standard deviation of circumference and weight as the objective function and the mean of draw resistance as the constraint condition, the multi-objective optimization design of process parameters was established based on the quadratic response surface, and the quality control mode of cigarette manufacturing quality combining these two aspects was established, and self-learning optimization and genetic algorithm were used to optimize the model. The ZJ19 cigarette making machine of Luohe Cigarette Factory was selected for online application analysis and verification of the cigarette manufacturing process of a certain brand. The results showed that the deviation between the average value of cigarette weight, circumference, draw resistance and the target value was reduced by 66.6%, 86.7%, 65.6% respectively, and the standard deviation was reduced by 31.8%, 65.2%, 11.9% respectively. This quality control mode combining deviation prediction adjustment and multi-objective parameter optimization can effectively improve the comprehensive quality level of the cigarette making process.